The new 3900X is underperforming in the R20 segment.
The new 3900X is underperforming in the R20 segment.
I changed from a Threadripper 1950X to a Ryzen 9 3900X and the scores matched closely on CPU Monkey. Now I have a 3600 MHz memory, but the result dropped to 460, which is 60 points lower than the public benchmark. How typical is this variation? Did adjusting my RAM speed affect the outcome? Raising the overclocking settings on the motherboard lowered the score too. I'm using the standard AMD Wraith cooler and an ASUS TUF X570 board with a fresh Windows 10 install. Any advice?
Restart and attempt once more. Occasionally I see low values, so it helps to run three times with a reboot in between. Right after reading your message I completed three runs: the first was 454, the second 485, and the last 500. These results aren’t very consistent. The multi-core scores suggest I’m performing better than expected—around 3833, which falls between R 5 3600x and R 5 3600xt.
I installed Ryzen Master, enabled the default precision boost overdrive, and achieved scores up to 484—about 24 points higher than my initial result. After a restart, the improvement wasn’t significant. I’m curious about how these scores translate in real life... what kind of projects do you usually work on? I mainly use Premiere for video editing.
I rely on my PCs just for gaming. As a former freelance 3D artist and designer, my machines have always been my main source of income until 2018. I believe strong scores don’t always reflect real performance in other areas. My i7 8086k model supports a 5GHz all-core overclock, while the other runs at 5.1GHz all-core. The only consistent win is Cinebench, with no noticeable improvement elsewhere. The visual results look good in screenshots, but I focus on single-core speed since I enjoy editing and building games. Stronger cores let me add more content. Some games become unplayable without my overclocking, and I can’t afford to upgrade until better CPUs appear. I used to prefer affordable high-core CPUs for low-budget jobs where rendering times mattered most. I never overclocked because a crash could ruin my work the next day. Any small boost from overclocking didn’t justify the risk.
It helps me understand why I enjoy being an editor because it gives me the chance to upgrade my hardware. Oddly enough, I turned off all MB overclocking, set my RAM XMP to 3600, removed Ryzen Master, and achieved a 497 score—my best so far without any overclocking. When I re-enabled precision boost overdrive, it jumped to 484, which is 24 points higher than my first result. After restarting, the improvement wasn’t huge. I’m curious about how these scores translate in real life and what you typically use your computer for—mostly video editing with Premiere.